There is a version of this story where a two-month-old DSP with eight customers gets marched into the town square for the crime of having a homepage. That is not this story. Doug Lauretano did something almost nobody in ad tech does when a reporter shows up with six pointed questions and a same-day deadline: he answered all of them, in writing, on the record, including the one where the truth made him look a little sloppy. In an industry built on dodged calls, ghosted emails, and the eternal "no comment," that's not just good manners. It's rarer than a clean bill of health from an ad tech verification company auditing itself.

And look, we've all been there. Anyone who has ever launched anything, a company, a newsletter, a lemonade stand with a hand-painted sign that misspells "refreshing," knows that the gap between the pitch deck and the actual plumbing is where every founder lives for the first year. Doug is two months into building a DSP from scratch. Give the man a beat.

So let's give him credit, and then let's still do the math. Because the math is where this gets fun.

Tuple's origin story starts with a real number, and a real hole in the floor. While running the advertising business at CivicScience, Lauretano says his team would hand SSPs 10 million impressions a day in a custom data package, and watch DSPs report seeing just 4 million of them. Not a glitch. Not one confused buyer. Consistent, every time, across the board, like a magic trick nobody clapped for because nobody realized six million doves had vanished behind the curtain.

QPS throttling and traffic shaping, the industry's polite names for "we didn't want to pay for the infrastructure to show you everything," were quietly composting perfectly good, fully identified, non-cookieless impressions on the cutting room floor. He puts the industry-wide toll at roughly 40% of unique impressions never reaching a buyer's DSP at all, and the 40% that does survive the trip shows up duplicated dozens of times over, the header-bidding equivalent of a photocopier stuck on repeat while everyone in the room insists it's producing forty different documents.

Tuple's answer to that mess: stop connecting to hundreds of SSPs and trying to read tea leaves in the fog. Connect to a small number, run decisioning inside a shipping container welded directly to the SSP's loading dock, deduplicate before anything crosses the threshold, and call the result unduplicated access.

Right now that small number of trusted supply partners is, arithmetically speaking, one. Media.net. A DSP built on the promise of diversified, verifiable supply, currently running on a single, solitary, entirely-alone-at-the-prom pipe.

One company. Three different counts.

Public materials: one SSP, Media.net, with an ambition to eventually reach three.

On a conference stage, two months ago, present tense, no hedge: "We're only working with two to three SSPs."

In his written reply to ADOTAT, today, under an actual deadline with the actual numbers on the desk: "We currently only have supply via Media.net... a second integration in progress and another under discussion."

That last one is the true one. It's the one that had to survive an inbox. One live SSP. Today. Full stop. Everything else is roadmap wearing a present-tense costume to a party it hasn't been invited to yet.

Not a smoking gun. Just a founder describing his own toddler as "practically fluent in three languages." Every parent does this. Every founder does this. But words are the whole product being sold here. "Absolute transparency" sits on the homepage like a neon sign over a diner that's still deciding on the menu. A company selling verifiability as its entire personality doesn't get to be casual about its own supply count, even sweetly, even understandably.

Credit where it's due: Lauretano didn't send a spokesperson, didn't send three careful sentences vetted by three careful lawyers, and didn't flinch on the two questions most likely to leave a mark.

On the preferred-partner risk that comes with betting an entire company on one SSP, the way a tightrope walker bets everything on one wire, he didn't retreat behind a policy statement. He handed over the receipts: "We welcome audits and/or the analysis of log level data, because unlike legacy DSPs, we have nothing to hide." That's a sentence most platforms ten times Tuple's size would need a full legal review, a risk committee, and possibly a burnt offering before they'd let it anywhere near a reporter.

On whether Tuple's own privacy policy still has unfilled placeholder text sitting where its California consumer-rights disclosures should be, next to a contact address that rings the digital equivalent of a disconnected phone number, he didn't pretend the house was tidy: "Yes, that is a known issue and should be corrected shortly. We also have one other placeholder in the privacy policy that will also be addressed soon." He then volunteered a second problem we hadn't even found ourselves, like a suspect walking into the station to confess to a crime the police hadn't gotten around to investigating yet. Companies caught with their pants down don't usually point out the second pair of pants. Most companies wouldn't have noticed the placeholder text at all until a reporter did. He'd already clocked it himself. That's the whole ballgame, honestly.

On the harder measurement question, whether any customer has run a real, independent, outside-the-family lift test against an incumbent DSP rather than letting Tuple grade its own homework, the answer is not yet, with case studies "hoped for" in Q4. Rather than dressing that up, he handed advertisers a genuinely useful DIY diagnostic instead: pull the avails an SSP reports on a Deal ID over seven days, compare it to what actually shows up in your own DSP for the same window, and watch the gap reveal itself like a tide going out on a beach nobody realized was this littered. That's a real test. It is not, and he didn't pretend it was, the causal, third-party comparison the question actually asked for.

None of that erases the concentration risk. It does mean the risk is currently being run by someone willing to show his math on the chalkboard instead of erasing it the second the bell rings.

Here's the part of this story that's easy to miss if you only read it for the receipts: a company this size has almost nothing to gain from lying to you, and everything to lose from getting caught.

A decade-old DSP with enterprise contracts, a sales floor, and a board expecting quarterly growth has every incentive to defend its black box the way a magician defends the trapdoor, because the trapdoor is already load-bearing revenue and admitting it exists threatens the whole show. Tuple, at eight customers and two months old, doesn't have a black box worth defending yet. The only asset on the balance sheet is whether buyers believe what it says. That is a genuinely different incentive structure than the one that produced the industry-wide opacity everybody complains about at every conference, on every panel, over every hallway coffee that's gone cold while people vent about SSPs.

Is that a permanent advantage? No. The day Tuple has real revenue worth protecting is the day this incentive starts to corrode, the way it corroded for every SSP and DSP that also used to talk a big game about transparency back when they, too, were small and had nothing else to sell but their word. But right now, today, a founder answering all six hard questions, including the embarrassing one, is doing something structurally rational, not merely personally admirable. Worth noting. Worth watching closely to see if it survives the first real taste of success, the way you'd watch to see if a New Year's resolution survives February.

And honestly? He's two months into this. Building the plane while flying it, filling out the privacy policy while filling out the customer roster, answering ADOTAT's inbox at 6pm on the same day he's presumably also trying to fix the SSP count problem this very article is about. That's not an excuse. It's just the actual texture of doing this, the part nobody puts on the pitch deck slide.

The concentration risk doesn't evaporate just because the founder answers email politely. One live SSP means one company's routing changes, fee structure, or QPS allocation can reshape Tuple's entire product overnight, and Tuple's customers would have no alternate path idling in the driveway. "100% visibility into Media.net's bidstream" is a genuinely strong claim, but it's visibility into one supplier's bidstream, the view from a single window, not a survey of the whole street. Self-reported dedup logic, however elegantly engineered, is not the same thing as an outside party independently confirming the arithmetic.

The fair ask, the one any advertiser should actually make before moving real budget: get the supply-coverage audit, get the path-level economics, get the causal lift test once one actually exists, and treat "we welcome audits" as an invitation worth RSVPing to, not a slogan worth admiring from across the room.

Zoom out, and there's a bigger, kinder thesis buried under all this arithmetic. Agencies used to sell media strategy and trading discipline, the craft of translating a client's actual intent into disciplined execution with proof it actually happened. A decade of holding-company consolidation and procurement-driven margin squeeze hollowed most of that out from the inside, like termites working a beam nobody thought to inspect. What's left in a lot of shops is contract management wearing a media-buying costume to the office party.

Meanwhile, the DSP layer, at least the part of it trying to compete on constraint-enforcement and auditability instead of just cheaper CPMs, is quietly absorbing the exact function agencies used to claim as their whole reason for existing. If a platform can enforce "never buy X," produce the log proving the constraint held, and hand a buyer the receipts, it's doing the job a trading desk used to promise over steak dinners and rarely got independently checked on.

That doesn't make Tuple, or anyone else building this way, "the new agency." A DSP doesn't do integrated planning, doesn't do creative, doesn't carry the client relationship through a crisis. It makes Tuple a candidate for something narrower and arguably more useful: the new trading desk, for buyers smart enough to keep strategy in-house and just want the execution layer to stop lying to their face. That's not a new agency.

That's the ghost of the old agency's best department, the one that already died once, looking for someone honest enough to hold a séance.